In March 2025, Tata Sons' own annual report told investors the Dholera semiconductor fab would begin its chip journey at the 28nm process node. Group chairman Natarajan Chandrasekaran said as much in writing. Union IT minister Ashwini Vaishnaw told reporters the plant would produce its first chip by December 2026. Neither of those things is now true. India's first large-scale chip fab is opening on 90nm technology — a node roughly two full generations behind 28nm — and Vaishnaw himself now says commercial production is expected in mid-2028, an 18-month slip from what he announced. Tata Electronics' current position is that this was the plan all along. That claim doesn't square with what the company's own chairman told shareholders in writing sixteen months ago.
What a "process node" actually buys you
The node number roughly describes the size of the transistors on a chip, and it is the single biggest lever on what that chip is good for. Smaller nodes pack more transistors into the same area, run faster per watt, and are what modern smartphone processors, AI accelerators and high-end computing chips are built on — that is the territory 28nm sits toward, relative to 90nm. A 90nm chip, by contrast, is mature, well-understood, cheap-to-yield technology, well suited to power management chips, automotive electronics, and the kind of unglamorous silicon that goes into a car's dashboard or an appliance's control board rather than a flagship phone. Both are real, useful chips. They are not remotely the same pitch to the country, or to the fab's future customers.
| Original pitch (Mar 2025 – early 2026) | Current plan (Aug 2026) | |
|---|---|---|
| Starting node | 28nm | 90nm, then 55nm, with 28nm later |
| First chip / silicon | December 2026 | Trial production Dec 2026, commercial mid-2028 |
| Stated by | Tata Sons annual report; N. Chandrasekaran | Ashwini Vaishnaw; Tata Electronics |
| What it targets | Advanced compute-class chips | Power management, automotive, telecom, defence electronics |
The company's explanation doesn't match the company's own paper trail
Tata Electronics' current line is that the fab was always designed to start at 55nm and 90nm before working up to 28nm, and that 28nm "will be a key part of our offering" further down the road. Read alone, that is a defensible engineering sequence — plenty of fabs bring up mature nodes first to prove out yield and process control before risking a newer, harder node on a production line. The problem is that it directly contradicts what Tata Sons told its own shareholders in the fiscal-2025 annual report, where Chandrasekaran described the fab's chip journey as starting at 28nm, not ending there. One of those two statements about the same project, from the same company, describing the same starting point, is wrong. Tata Electronics has not addressed the contradiction directly — only restated its current version as though it were the only version there has ever been. Taiwan's Powerchip Semiconductor Manufacturing Corp, Tata's technology partner on the project, has described the tie-up as covering multiple nodes with 28nm as the most advanced — consistent with the current story, not the original one.
Why the timeline slipped is a separate, more ordinary story
None of this is unusual for first-of-its-kind fabs. Standing up a country's first advanced semiconductor plant from nothing — the clean rooms, the ultra-pure water and gas supply chains, the workforce trained on equipment India has never operated at this scale before — is a genuinely hard industrial problem, and slipped timelines are closer to the global norm than the exception; Intel, Samsung and TSMC have all pushed announced node and facility timelines by a year or more on new sites. Trial production in December 2026, roughly matching the original target for some silicon to come off the line, followed by a slower ramp to commercial volumes through 2027 and into mid-2028, reads as a fab that hit real engineering friction and chose to de-risk by proving out an easier node first, rather than one that has quietly failed. That is a legitimate, common decision.
Dholera isn't the only semiconductor project running behind India's headline dates
It's worth separating what's actually operating today from what's still a promise. Micron's assembly-and-test facility in Sanand, Gujarat — a genuinely different, easier step in the chip supply chain than building a fab from raw silicon wafers — was inaugurated by the Prime Minister in February and is on track to assemble and test tens of millions of chips through 2026, scaling toward hundreds of millions in 2027. That's real, running output, and it's the part of the India Semiconductor Mission that's actually ahead of schedule. Fabrication — growing the silicon wafer itself, which is what Dholera does — is the far harder, far more capital-intensive link in the chain, and it's the one where India has no prior experience to draw on at all. Conflating the two, as some coverage of the mission tends to do, makes the overall programme look further along than the fabrication piece specifically is. Dholera slipping while Sanand ships on schedule isn't a contradiction — it's the predictable gap between the easy end of a supply chain and the hard one.
What actually got sold, and to whom
The distinction that matters is between the engineering decision and the public pitch. Starting a first fab on a mature node to protect yield is a sound choice a company can make on its own judgement. Telling shareholders and the government the plant would begin at 28nm, then telling the public a year later that 90nm-first was the plan from day one, is not an engineering decision — it's a communications one, and it lands differently. The ₹91,000-crore Dholera project was announced, funded and incentivised in significant part as India's entry into cutting-edge chip manufacturing, the kind of capability that reduces reliance on Taiwan and South Korea for the chips inside phones, laptops and AI hardware. A fab that opens on 90nm and scales toward 28nm "later" is still a genuine, valuable industrial milestone for India — its first large-scale semiconductor wafer facility, full stop. It is just not the milestone that was originally described to the people funding and championing it, and the gap between those two versions is worth more scrutiny than Tata Electronics' current framing is giving it.
